Anjorin, I. (2009). High-stakes tests for students with specific learning disabilities: Disability-based differential item functioning (Publication No. 3390841) [Doctoral dissertation, Southern Illinois University at Carbondale]. ProQuest Dissertations and Theses Global. https://www.proquest.com/docview/304997473
Southern Illinois University (Carbondale, IL); ProQuest document ID: 304997473
Thirty-one types of accommodations were made available to students taking the 10th grade math test. Accommodations were given either individually or in packages/combinations as identified by each student's IEP. The most common packages were: small group/read aloud directions/sign interpret directions/clarify directions; small group/calculator; calculator/ number chart/manipulatives. The most frequent single accommodation (not in a package) was small group administration.
An extant dataset of statewide test performance scores was examined. There were data for 55,313 students without disabilities, and 5,957 students with specific learning disabilities (SLD). Participants were divided into four groups (one reference group and three focal groups): students without disabilities (reference), all students with SLD, students with SLD who received accommodations on state assessments, and students with SLD who received no test accommodations.
Student performance on items from a grade 10 mathematics assessment in one unidentified state (U.S.) was used to calculate differential item functioning (DIF).
Descriptive statistics revealed that mean scores for students without disabilities (SWOD) were 10+ points higher than for students with SLD. Items were then examined for differential item functioning (DIF). When examining the level of DIF in performance between SWOD and students with SLD, 62% of the items exhibited no DIF, and 38% of items exhibited variable DIF favoring either group (19% favored SWOD, 19% favored students with SLD). The items that favored students with SLD were most often items on which they could use a calculator (one of the most commonly-provided accommodations). Items that favored students with SLD were also found to be more difficult to compute. More DIF items were found to favor the SWOD group, which is potentially concerning, and an area that the author identified as needing further research.